llm2vec_config.json

{
    "pooling_mode": "last_token",
    "max_length": 1024,
    "doc_max_length": 400,
    "skip_instruction": true
}